Thread: pg_dump outputs invalid syntax for partitions with bool partkeys in pg10

Quite simply:

d1=# create table bp (b bool) partition by list(b);
CREATE TABLE
d1=# create table bp_f partition of bp for values in('f');
CREATE TABLE
d1=# \q
$ createdb d2
$ pg_dump d1 | psql d2

...

ERROR:  syntax error at or near "false"
LINE 2: FOR VALUES IN (false);
                       ^
ERROR:  relation "public.bp_f" does not exist
ERROR:  relation "public.bp_f" does not exist
invalid command \.

I understand there's a thread [1] which is discussing making this
valid syntax, but on skimming it, there's no mention of any bugs.

I'd thought about something like the attached patch (against master)
to fix, but that leaves pg_dumps that have already been taken a bit
out in the cold, so perhaps we need to think harder about allowing
this syntax.

Thoughts?

This was noticed by Jesper Pedersen in [2], but I've diagnosed this as
an existing bug rather than a bug in the patch that thread relates to.

[1]
https://www.postgresql.org/message-id/flat/e05c5162-1103-7e37-d1ab-6de3e0afaf70@lab.ntt.co.jp#e05c5162-1103-7e37-d1ab-6de3e0afaf70@lab.ntt.co.jp

[2] https://www.postgresql.org/message-id/c3eb3284-24c8-eff0-f930-a33984d1168a@redhat.com

-- 
 David Rowley                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services

Hi David.

On 2018/03/02 12:41, David Rowley wrote:
> Quite simply:
> 
> d1=# create table bp (b bool) partition by list(b);
> CREATE TABLE
> d1=# create table bp_f partition of bp for values in('f');
> CREATE TABLE
> d1=# \q
> $ createdb d2
> $ pg_dump d1 | psql d2
> 
> ...
> 
> ERROR:  syntax error at or near "false"
> LINE 2: FOR VALUES IN (false);
>                        ^
> ERROR:  relation "public.bp_f" does not exist
> ERROR:  relation "public.bp_f" does not exist
> invalid command \.
> 
> I understand there's a thread [1] which is discussing making this
> valid syntax, but on skimming it, there's no mention of any bugs.

Thanks for the report.

I guess we don't have a Boolean partitioned table in tests covered by
pg_dump or I'd have heard about this syntax glitch before Horiguchi-san
mentioned it a few months back [1].

> I'd thought about something like the attached patch (against master)
> to fix, but that leaves pg_dumps that have already been taken a bit
> out in the cold, so perhaps we need to think harder about allowing
> this syntax.

I agree about going ahead with adding the syntax support, but there are
some arguments against the design of the proposed patches or rather about
the original design of partition bound parsing itself.

> This was noticed by Jesper Pedersen in [2], but I've diagnosed this as
> an existing bug rather than a bug in the patch that thread relates to.

Certainly.

Thanks,
Amit

[1]
https://www.postgresql.org/message-id/20171128.203915.26713586.horiguchi.kyotaro%40lab.ntt.co.jp